Deep Learning for Photovoltaic Generation Forecast in Active Solar Trackers

Abstract

The generation of electricity by photovoltaic panels depends on the position of solar incidence on them. Using active solar trackers may be a maximization of generating capacity. However, if motors that update the position of the panels use more energy than the efficiency in their use, the system becomes ineffective. In this way, solar forecasting can be used to actively determine the generation capacity and to assess whether position updating is efficient. Among the algorithms that can be used to predict photovoltaic generation, stands out the Long Short-Term Memory (LSTM) which is an artificial recurrent neural network architecture used in deep learning. This technique stands out among the others for having the ability to handle complex problems with high nonlinearity. The results of the application of LSTM for photovoltaic generation forecast in active solar trackers are promising as described in this article.